Contents
- What Meta Ads are good at, and what they are not
- Business Manager and asset ownership
- The Pixel and the Conversions API
- Choosing the objective
- Audiences: who you actually reach
- Campaign structure and budgets
- Creative, which is most of the result
- Lead forms, WhatsApp or a landing page
- Policy, restrictions and keeping the account
- Measuring and optimizing
- What it costs
- Mistakes that waste most of the budget
What Meta Ads are good at, and what they are not
Facebook and Instagram ads interrupt people. Nobody on Facebook is searching for a dentist, a trekking package or a study-abroad consultancy at the moment your ad appears. That is the whole difference from Google Ads, and it decides everything else: your creative has to earn the attention, your offer has to be easy to say yes to, and your expectations of a first-click sale should be low for anything expensive or considered.
What Meta does better than any other channel available in Nepal is reach at low cost, precise retargeting of people who already showed interest, and lead generation for services people buy emotionally or urgently: clinics, gyms, salons, events, schools, property, travel. It is also where most Nepali consumers already spend their time, which is why a business with no website at all can still generate enquiries here.
What it is bad at: capturing existing demand. If someone is searching “emergency plumber Baneshwor” right now, you want Google Ads. Most businesses eventually run both, and the trade-off is set out in Facebook Ads versus Google Ads.
Business Manager and asset ownership
Start with a Meta Business Manager owned by the business, with the owner as admin. Then bring the Facebook page, Instagram account, ad account, Pixel or dataset, catalogue and WhatsApp number into it as business assets. This sounds administrative until the day an agency relationship ends, a staff member leaves, or a personal profile is disabled and takes the page with it. In Nepal I regularly meet businesses whose page is owned by a former intern and whose ad account is on someone’s personal profile with a personal card attached.
Practical rules: two-factor authentication for everyone with access; give partners the least access they need; verify your domain in Business Manager; keep a business payment method with a backup, because a declined card pauses everything overnight. Then complete the page itself. An ad running from a page with no photos, no posts and no contact details converts worse and attracts more scrutiny.
The Pixel and the Conversions API
Install the Pixel once on every page of the site, and fire standard events for real actions: Lead for an enquiry, Contact for a call or message, Purchase with a value and currency for a sale. Then add the Conversions API alongside it, with deduplication configured, so conversions still reach Meta when the browser blocks the Pixel. Since browser tracking became unreliable, an account running on the Pixel alone is optimizing on partial data and usually reports fewer results than it produced.
Test every event yourself in Events Manager before you spend anything. Send the parameters you legitimately can, hashed, because match quality directly affects how well delivery works. Do not send anything sensitive: for a clinic, record that an enquiry happened, not what it was about. This is the work covered by the Meta Pixel and Conversions API service, and the full list of checks is in the Facebook Ads checklist.
Choosing the objective
The objective tells Meta which people to find, so it matters more than any targeting choice you make afterwards. Choose Leads if you want enquiries, Sales if you want purchases and have the events to prove them, Engagement only when you genuinely want comments or messages, Traffic only when a visit is the outcome you want, and Awareness only when you are deliberately paying for reach.
The common failure is choosing Traffic because clicks look cheap. Traffic optimization finds people who click and do nothing else, and the report looks excellent while the phone stays quiet. If you have no conversion data at all, it is better to start with Leads using an instant form than to buy clicks you cannot judge.
Boosting a post from the page is not a substitute. It gives you almost none of the controls in this guide, which is why I treat it as a separate, limited service on the Facebook boost service page rather than as advertising.
Audiences: who you actually reach
Meta’s delivery system is better at finding buyers than most manual targeting, provided you give it a signal to work from. In practice, three layers do the work.
- Your own data. Website visitors by page, video viewers, page and profile engagers, form openers, and a customer list you have permission to upload. These are the cheapest conversions you will ever buy, and they are the basis for retargeting.
- Lookalikes built from the best source you have, which is customers rather than all visitors. A lookalike from 200 real buyers beats one from 20,000 page viewers.
- Broad or lightly interest-targeted audiences for acquisition, where the creative does the targeting. In a market the size of Nepal, stacking five interests and three behaviours usually leaves the audience too small to optimize.
Set exclusions so existing customers and current leads stop seeing acquisition ads, use “people living in this location” unless you want travellers, and check audience overlap so your ad sets are not bidding against each other.
Campaign structure and budgets
Keep it simpler than you want to. Meta needs roughly 50 optimization events per ad set per week to leave the learning phase and deliver predictably. Splitting a modest Nepali budget across eight ad sets guarantees that none of them ever gets there. One acquisition campaign with one or two ad sets and three or four creatives, plus a separate retargeting campaign, is the right shape for most businesses here.
Separate acquisition from retargeting so their very different costs per result do not average into a number that means nothing. Name campaigns so you can read them in six months. Leave placements on Advantage+ unless you have measured a reason not to, and supply the creative in the ratios each placement needs. If you use Advantage+ campaigns, know which controls you have handed over and set the existing-customer budget cap deliberately.
Change one thing at a time and leave it alone for three or four days. Every meaningful edit restarts learning, and an account edited daily never gets out of it.
Creative, which is most of the result
On Meta the creative is the targeting. The same offer, the same audience and the same budget will produce wildly different costs per lead depending on the first three seconds of the video. What works consistently in Nepal:
- A real person speaking to camera in the language the audience speaks, usually Nepali, usually the owner, doctor, trainer or specialist rather than an actor.
- Vertical video for Reels and Stories, with captions burned in, because most people watch without sound.
- The subject in the first three seconds. No logo animation, no slow build.
- Real premises, real staff, real work. Stock photography is recognised instantly and ignored.
- Three genuinely different concepts per ad set, not three colours of the same graphic, so you learn which idea works rather than which shade of blue.
- A refresh schedule. On a local audience, frequency climbs fast and the same creative stops working within weeks.
Production and concepts are the Meta ad creative service, and the format-specific detail sits on the video ads page.
Lead forms, WhatsApp or a landing page
Three destinations, three different trades.
Instant lead forms keep the person inside the app and produce the most leads per rupee, at the cost of quality. Use the higher-intent setting and add one qualifying question, or you will pay for people who tapped by accident. This is the route in lead generation ads.
WhatsApp or Messenger suits Nepal well, because a large share of customers would rather message than fill in a form or answer a call from an unknown number. It works only if someone replies within minutes. See WhatsApp and Messenger ads.
A landing page gives you the space to answer objections and to retarget the visitor afterwards, and it is the right choice for a considered purchase. It has to load in under three seconds on mobile data and repeat the ad’s promise in its headline; the checks are in the landing page checklist.
Whichever you choose, response speed decides your real cost per customer. A lead answered in five minutes and a lead answered the next morning are not the same lead.
Policy, restrictions and keeping the account
Meta enforces policy automatically first and reviews later, so the practical goal is to avoid the trigger. Health, finance, employment, housing, credit and anything to do with personal attributes carry extra restrictions. Do not write copy that implies you know something personal about the reader (“struggling with infertility?”), do not use before-and-after body imagery, and do not promise outcomes you cannot support.
Declare special ad categories where they apply and accept the targeting limits that come with them. Keep your business verification current. Assume a restriction will happen at some point and make sure more than one person has admin access when it does. Recovery work is the account restriction and policy compliance service.
Measuring and optimizing
Judge the account on qualified enquiries and customers, not on reach, engagement or the number Meta reports for leads. Reconcile Meta’s reported results against the inbox weekly; where they disagree, the inbox is right. Watch frequency and cost per result together, because a rising cost per result with a rising frequency is creative fatigue and needs new creative, not new targeting.
Tag every link with UTM parameters so analytics can attribute traffic independently, and expect Meta’s attribution and GA4’s to disagree; they count differently, and the useful question is whether total enquiries rose while spend was flat. The measurement setup is in the GA4 and GTM setup checklist.
What it costs
Two costs: the ad spend and, if you hire someone, the management fee. For Nepal campaigns, ad spend of NPR 15,000 to 30,000 a month is the level at which results become readable; below that, weekly swings are larger than the effect of anything you change. International campaigns generally need USD or AUD 500 to 1,000 a month. My own management fees are Starter at USD 500 a month for a simple account, Growth at USD 900 a month, and Enterprise quoted after scoping for complex multi-funnel accounts; international clients are invoiced in their own currency at rates 40 to 60 percent below comparable Western agency rates.
Cost per lead varies so much by industry, offer and creative that any single number quoted as a benchmark is misleading. What I can give you is one real example: a Kathmandu fertility clinic where doctor-led Nepali video and mobile lead forms produced more than 150 qualified leads at about NPR 180 per lead. That is one campaign in one industry, not a rate card. The wider picture is in the digital marketing cost guide for Nepal.
Mistakes that waste most of the budget
- Boosting posts instead of running campaigns, and calling the result advertising.
- Optimizing for Traffic or Engagement when you want enquiries.
- Ten ad sets on a small budget, all permanently in learning.
- One creative, run until it dies, then blamed on the algorithm.
- English creative for an audience that thinks and buys in Nepali.
- Lead forms nobody answers for a day and a half.
- No Conversions API, so half the conversions are invisible to the system that is meant to find more of them.
- An ad account and page owned by someone who no longer works with you.
If you would rather have this run for you, the services are Meta Ads management and Facebook and Instagram advertising generally, and a free Meta Ads audit will tell you which of the mistakes above are currently costing you money. The step-by-step version of this guide is the Facebook Ads checklist; everything else is on the resources page.
Last updated 2026-09-03.
Updated